Recognizing primary HIV infection (PHI) is important for patients and for public health. During this initial phase of HIV infection, HIV-specific CD4 T lymphocytes that best respond to HIV may be destroyed by the virus, permanently impairing the immune system's ability to control HIV . Treatment of PHI may offer unique opportunities to preserve HIV-specific immune responses . PHI may also represent an important opportunity to interrupt HIV transmission because persons in this stage of HIV can be an important source of new infections . Persons in PHI have high HIV-1 RNA levels prior to the development of effective immune responses, which probably increases infectiousness . They are also typically unaware of having HIV infection and may be in a period in which they are engaging in risk behavior that might transmit HIV to others.
It is difficult to identify persons with a substantial probability of having primary HIV infection when evaluating risk behaviors alone. Even the highest risk exposures, such as receptive anal intercourse or sharing drug injection equipment with an HIV-infected partner, are thought to have a risk of HIV transmission less than 3% from a single exposure . Many exposures are lower risk, such as vaginal or insertive anal intercourse, or exposures with partners of unknown HIV status. Appropriate selection of persons for testing for primary HIV infection is important because diagnostic tests for this condition have limitations in accuracy, errors in diagnosis can be psychologically distressing, and tests can be expensive. Early in infection, HIV-1 antibody tests are unreactive, and diagnosis relies on p24 antigen or HIV-1 RNA tests. While p24 antigen testing is less expensive and probably has better specificity, there may be limitations in its sensitivity. HIV-1 RNA testing is more sensitive but may have a higher risk of false-positive tests and is more expensive.
Up to 90% of persons who become HIV infected experience an initial influenza-like illness that typically occurs between 1 and 4 weeks after infection . This clinical condition has been called acute retroviral syndrome . Its symptoms have been well described in case series and in cohort studies following persons at high risk of HIV infection [8,9]. However, little is known about the utility of these symptoms in a clinical setting for distinguishing persons with suspected PHI who are actually infected from those who are not. For the health-care provider who must decide whether to consider a diagnosis of PHI in a patient presenting for care, specificity becomes particularly important in avoiding unnecessary testing of uninfected persons.
The present study has evaluated the diagnostic characteristics of symptoms within a study that evaluated persons with suspected PHI in order to describe the utility of specific symptoms in distinguishing persons with PHI from those without. These characteristics can be useful in determining who should receive laboratory testing for PHI. The performance of the following laboratory tests were then evaluated for the diagnosis of PHI: a p24 antigen EIA, a third-generation HIV-1 antibody test capable detecting IgM anti-HIV-1 antibodies, which is routinely used in blood donor screening, and three types of HIV-1 RNA test.
The University of California at San Francisco (UCSF) Options Project recruited subjects with potential primary HIV infection who met one of two criteria: potential acute retroviral syndrome and potential recent HIV antibody seroconversion. The first group, those with potential acute retroviral syndrome, comprised persons with a possible sexual, drug use, or occupation exposure to HIV in the prior 3 months and at least one of the following symptoms: fever, rash, pharyngitis, arthralgias, night sweats, or lymphadenopathy. Persons with unprotected receptive anal sex or shared drug use equipment with a partner known to have HIV infection in the past 3 months could also be screened in this group even if symptoms were not present, because of the high risk of infection with these exposures. The second group, those with potential recent HIV antibody seroconversion, comprised persons with a negative HIV-1 antibody test within the past year and a recent positive antibody test or, if the last HIV-1 test was more that 12 months ago and was negative, there were risks of HIV exposure in the past 6 months. To be screened, potential participants also had to be 18 years of age or older, or at least 16 years old if an emancipated minor. Potential participants were excluded if they had taken prior antiretroviral treatment. The study was approved by the Institutional Review Board of the UCSF and participants had to sign informed consent prior to the screening process.
Participants for screening could be self-referred or could be referred by others, including health-care workers, HIV testing sites or community-based organizations. A variety of methods were used to publicize the study and recruit participants. These methods included outreach to health-care workers through speaking events, mailings, and word of mouth, providing referral information to staff at local HIV information hotlines, encouraging referrals from local HIV testing sites, and establishing referrals from research cohorts that might detect recent HIV seroconverters.
To increase the number of persons with preseroconversion PHI for assessing the performance of laboratory tests for PHI, preseroconversion specimens from seven patients from a cohort in Boston were also tested. The cohort from which these additional specimens were obtained have been described elsewhere .
Symptoms were assessed using a standardized questionnaire that was self-administered for the first 178 (44%) patients, and interviewer-administered thereafter. Participants were asked whether they had recently experienced any of the 21 symptoms: fever, rash, oral ulcers (mouth sores), arthralgias (Joint pain), pharyngitis (sore throat), loss of appetite, lost weight (> 5 lb; 2.5 kg), malaise (felt sick), myalgias (pain in muscles), tired or fatigued, nausea, headaches, photophobia, night sweats, confusion, infected gums, diarrhea, sores on genitals, vomiting, sores on anus, and stiff neck. For each symptom experienced, participants were asked start and stop dates. In order to assess the symptoms that would be considered by a physician, symptom reports were reviewed by one of two investigators (JA, FH). Symptoms were excluded if they had lasted more than 4 months as these were assumed to be chronic symptoms that preceded HIV infection. Symptoms were also excluded if they began more than 4 weeks before or after a cluster of symptoms being considered as potential acute retroviral syndrome. If both investigators agreed there was a strongly possible alternative etiology for the symptom, such as genital ulcers consistent with herpes simplex in a patient with a history of herpes simplex, the symptom was also excluded.
p24 antigen was measured using an enzyme immunosorbant (EIA) assay (Abbott, Abbott Park, Illinois, USA). To maximize sensitivity to early viremia, p24 antigen testing was performed without acid dissociation, which improves detection of p24 antigen when antibodies are present but makes the assay less sensitive to low levels of p24 antigenemia prior to the development of antibodies. Plasma HIV-1 RNA was measured using three different tests: the Bayer branched-chain DNA (bDNA) test version 2.0 (lower limit of detection 500 copies/ml) between June 1996 and July 1998, and test version 3.0 (lower limit of detection 50 copies/ml) between August 1998 and July 2000 (Bayer, Emeryville, California, USA); the Roche Amplicor polymerase chain reaction (PCR) (Roche Molecular Systems, Branchburg, New Jersey, USA); and a transcription-mediated amplification (TMA) HIV-1 RNA test (Gen-Probe, San Diego, California, USA), which is currently used to screen blood bank donations in the United States and provides results that are interpreted as either positive or negative . Standard HIV-1 EIA tests, the Abbott 3A11 EIA (Abbott) or the Vironostica HIV-1 EIA (Organon-Tecnika, Boxtel, the Netherlands), confirmed with a Western blot (Cambridge Biotech, Rockville, Maryland, USA). HIV-1 antibodies were also measured using a third-generation, recombinant HIV-1 antigen sandwich EIA assay (Combi test, 3A77; Abbott). Third-generation EIA HIV antibody tests are capable of detecting both IgG and IgM antibodies, whereas earlier antibody tests, which are used in most clinical testing, assess only IgG antibodies. As IgM antibodies usually appear before IgG antibodies, third-generation EIA HIV antibody tests can detect infection earlier than standard EIA antibody kits.
Participants were classified as preseroconversion HIV infected if (i) testing using standard EIA HIV-1 antibody testing and confirmatory Western blot was negative or indeterminate, and (ii) an HIV-1 RNA test or p24 antigen test showed viral antigen followed by subsequent antibody seroconversion, or HIV-1 antibody seroconversion occurred within the next 12 weeks without subsequent HIV exposures.
A combination of subject history and laboratory testing was used to classify participants with a reactive standard HIV-1 antibody test as postseroconversion recently infected. Current Centers for Disease Control and Prevention (CDC) criteria were used for interpreting a Western blot, which was considered as reactive if at least two of the three bands p24, gp41 and gp120/160 were present . To be classified as having postseroconversion recent infection, participants had to have a history of having had a negative HIV-1 antibody test within the past 3 years and recent HIV exposures, and a less-sensitive EIA (LS-EIA) test that was not reactive . The LS-EIA uses a modified EIA test that requires higher levels of antibody with greater binding avidity to test reactive and takes a median of 129 days longer to become reactive than a standard EIA test . Participants were classified as long-term HIV infected if their HIV-1 antibody test was reactive and their LS-EIA test was reactive.
To be classified as HIV uninfected, participants had to have a non-reactive HIV-1 EIA antibody at screening and at follow-up at 12 weeks. Antibody-negative patients with any HIV-1 RNA measure consistent with the presence of HIV-1 RNA at screening also had follow-up testing with at least one repeat HIV-1 RNA assay that did not show detectable HIV-1 RNA at the same time that an HIV-1 antibody test was non-reactive.
To analyze the role of symptoms in predicting primary HIV infection, participants were excluded if they had a known positive HIV antibody test at the time of initial study evaluation, in order to focus on participants in whom the diagnosis of HIV was uncertain. This also insured that symptoms were considered that prompted evaluation for PHI, as some of the participants with known HIV infection were evaluated because of recent seroconversion rather than PHI symptoms. Among the remaining participants, the symptoms were compared in two groups: (i) recently infected subjects, which included both the pre- and postseroconversion groups defined; and (ii) subjects who were not recently HIV infected. This second group included both patients without HIV infection and those with long-term HIV infection. The following groups of individuals were excluded from this analysis because the duration of HIV infection was uncertain: those with an LS-EIA absorbance 0.75–1.5 of a standard control, and those with a non-reactive LS-EIA but with a CD4 cell count of < 200 × 106 cells/l. The cut-off used for defining a non-reactive LS-EIA test was 0.75, but a LS-EIA absorbance of up to 1.5 may still indicate recent HIV infection. CD4 counts < 200 × 106 cells/l can cause falsely non-reactive LS-EIA tests in persons with advanced HIV infection .
For analysis of the performance of diagnostic tests for persons with preseroconversion primary HIV infection, two groups were compared: (i) those with preseroconversion primary HIV infection and (ii) HIV-uninfected participants.
Sensitivity and specificity were calculated using conventional definitions , and 95% confidence intervals (CI) were calculated using the binomial distribution . Statistical analysis was conducted using SAS version 7 software (Cary, North Carolina, USA). Multiple logistic regression models of the symptoms that predicted PHI were constructed entering all symptoms that were associated with this outcome with a P value < 0.2.
Between 1 June 1996 and 31 December 1999, the UCSF Options Project screened 406 people for PHI (Fig. 1). Of these, 118 were already known to be HIV infected at first evaluation and were excluded from further analysis; 105 of these participants had evidence of recent seroconversion based on a documented recent negative antibody test (n = 35) or LS-EIA testing (n = 70). Of the remaining 288 subjects, 40 had PHI: 22 of these were preseroconversion and 18 were recent HIV seroconvertors. There were 209 subjects who were uninfected and nine who were determined to have long-term HIV infection, based on a LS-EIA test with standardized absorbance of > 1.5. A group of 30 participants with a positive HIV antibody test were excluded because the duration of infection was uncertain. Of these 30 excluded participants, 21 had a LS-EIA in which the standardized absorbance was 0.75–1.5, six did not have a LS-EIA test performed and three had a non-reactive LS-EIA but a CD4 cell count < 200 × 106 cells/l.
The majority white race/ethnicity in San Francisco participants closely matches the local demographics of HIV: 75% of reported AIDS infections are in whites . The predominant mode of potential exposure for participants was male–male sexual contact (Table 1). This reflects the characteristics of the HIV epidemic in San Francisco, in which 90% of reported AIDS cases are in men who have sex with men .
Symptoms in diagnosis of primary HIV infection
To determine which symptoms were most strongly associated with PHI among those with suspected infection, the symptoms in the 40 persons with confirmed PHI were compared with those in the 218 subjects without HIV or with undiagnosed long-term infection. In bivariate analyses, the following symptoms were associated with the diagnosis of PHI among persons with suspected infection: fever, rash, oral ulcers, arthralgias, pharyngitis, loss of appetite, weight loss of more than 5 lb (2.5 kg), malaise and myalgias (Table 2). Of these symptoms, the most sensitive for the diagnosis of PHI were fever (80%) and malaise (68%). The most specific symptoms for PHI were weight loss (86%) and oral ulcers (85%). A series of other symptoms that are common features of PHI such as headache, diarrhea, and night sweats were also common in persons without PHI and were of little use in distinguishing persons with the diagnosis.
The average duration of most PHI symptoms was 7 to 10 days (Fig. 2). Genital ulcers tended to last longer in persons with PHI (27 days) than in persons without PHI (9.5 days) and this was a statistically significant difference. Other symptoms that had different durations in persons with PHI compared with those without were photophobia (7 days compared with 10 days) vomiting (2 days compared with 7 days) and fatigue (19 days compared with 14 days). The duration of other symptoms was similar in both groups.
The independent predictive power of individual symptoms were examined in a multiple logistic regression model. The best independent predictors of PHI were fever and rash (Table 3). Oral ulcers and pharyngitis were close to statistical significance as predictors of PHI.
Laboratory tests for primary HIV infection
The three tests for HIV-1 RNA were 100% sensitive for preseroconversion PHI (Table 4). However, bDNA testing for PHI had a 5% false-positive rate, PCR testing had a 3% false-positive rate and the transcription-mediated amplification assay had a 2% false- positive rate. The false-positive results on the bDNA test ranged from 584 to 2058 copies/ml. The false-positive PCR tests ranged from 58 to 103 copies/ml. In contrast, the lowest HIV-1 RNA level in a person with preseroconversion HIV was 2809 copies/ml using the bDNA test. The false-positive HIV-1 RNA tests occurred for different patients using different assays. When stored plasma from separate specimen tubes obtained at the same visit that initially generated false-positive HIV-RNA tests were retested on the same assay, all were found to have no detectable HIV-1 RNA, consistent with laboratory error as the basis of the false-positive results.
The p24 antigen test had specificity of 99.5%. The one individual with a false-positive test scored just above the cut-off value for a positive test and was negative when retested on a stored specimen. However, the p24 antigen test had a lower sensitivity than HIV-1 RNA testing: 79%. Among the six persons with PHI in whom the p24 antigen test was negative, the range of HIV-1 RNA (by bDNA testing) was 2809 to 315 600 copies/ml. All had a reactive antigen sandwich EIA antibody test, and four had a reactive standard HIV-1 EIA antibody test, but the Western blot was indeterminate.
The third-generation EIA HIV-1 antibody test had a sensitivity of 77% and a false-positive rate of 3%. All of the patients with a false-negative third-generation EIA antibody test had a positive p24 antigen test. The six persons with false-positive test results were all negative on HIV-1 RNA tests, and did not convert their standard HIV-1 antibody tests on follow-up.
The two symptoms that most effectively distinguished persons in our cohort with PHI from those without were rash and fever. Rash was one of the most specific symptoms for PHI. The combination of rash and fever had high specificity but limited sensitivity for PHI. Other symptoms that were useful in identifying persons with PHI were oral ulcers, arthralgia, pharyngitis, loss of appetite, weight loss, malaise, and myalgia.
These data can be used in combination with the potential risk of exposure to HIV to estimate pre-test probability of PHI as a basis on which to decide if diagnostic laboratory testing is warranted. Precise estimates of the risk of particular exposures are difficult to make because limitations in existing data give rise to substantial variation. Factors such as high partner viral load and the presence of other sexually transmitted diseases can substantially increase the per contact risk of HIV transmission, but these factors are frequently unknown and remain difficult to quantify accurately even when they are. Despite these uncertainties, data on the risk of specific HIV exposures allow stratification of patients into low-, medium- and high-risk groups for acquiring HIV infection [4,5,15–20]. High-risk exposures include receptive anal intercourse and shared injection drug use equipment; medium-risk exposures include vaginal intercourse; and low-risk exposures include receptive oral sex with male partners. Our data on symptoms can be used to improve the estimation of probability of having PHI in patients with both HIV exposure risks and symptoms compatible with PHI. The likelihood ratios can be used in combination with data on the risks of specific exposures to determine the pre-test probability of PHI. While more accurate estimates can be made using simple calculations or nomograms created for this purpose , the relatively low probability of HIV infection from a recent exposure means that multiplying the probability of HIV infection by the likelihood ratio for a particular symptom is a reasonable estimate of the probability of PHI. This potentially allows physicians to improve their selection of patients for laboratory screening for PHI.
A series of other symptoms that are commonly reported to be present in PHI, such as headache and diarrhea, were not effective in distinguishing persons with PHI from those without. This is probably because of the frequency with which these symptoms occur in persons with potential HIV exposure who are not infected. The duration of symptoms was not found to be useful in most cases for identifying persons with PHI, although oral and anal ulcers typically had a longer duration in PHI.
Most prior reports of symptoms in persons with PHI have reported only the frequency with which certain symptoms were present in case series. Our data are consistent with these case series in showing that symptoms such as fever, rash, and pharyngitis are common in PHI [6,8]. Several symptoms, such as weight loss and loss of appetite, were more common in our subjects with PHI. One difference may have been our use of a standardized questionnaire to determine whether these symptoms were present, rather than using medical record review.
Other studies have determined which symptoms distinguished persons with PHI in settings that differ from that of the current study. A study of 22 seroconvertors in a cohort study in Australia found that fatigue, fever, lymphadenopathy, night sweats and headaches distinguished persons with PHI from matched seronegative controls . Another study screened blood specimens of persons presenting at a sexually transmitted disease clinic in India for the presence of p24 antigen in the absence of HIV-1 antibodies . This study found that fever, joint pain and night sweats were useful in distinguishing persons with PHI. A cohort study in Africa found that recent experience of fever, vomiting, headache, diarrhea, arthralgia, myalgia, rash and adenopathy distinguished women who had acquired HIV infection during the last follow-up period from those who had not . Fever is consistently present in each of these studies. However, there are several important differences between these studies and our study. Most importantly, the patient populations differ in that these prior studies were not addressing persons who were presenting for clinical evaluation of potential PHI. Symptoms such as headache and night sweats were not useful in distinguishing individuals with PHI in our data. This is probably because of the frequency of these symptoms in persons presenting with other illnesses that may suggest PHI. For example, only 4% of HIV-negative controls in the study by Fox and colleagues in Australia reported experiencing headache, compared with 43% of those with suspected PHI who were not infected in our study. In addition, studies in the developing world may not be fully applicable in the developed world, as the frequency of symptoms from other common illnesses may differ.
Our study is most similar to a recent report by Daar and colleagues, which evaluated symptoms in 204 persons presenting for suspected PHI . Fever, myalgias, rash, night sweats, arthralgias and the absence of nasal congestion predicted PHI. These findings are similar to those in our study, with the exception that we did not find night sweats to be useful and we did not ask about nasal congestion. In addition, several of the symptoms we found to be useful, including weight loss and anorexia, were not included in the study by Daar, and we found oral ulcers to be more accurate in distinguishing persons with PHI.
p24 antigen testing was the first assay available for diagnosis of HIV infection prior to antibody seroconversion and has been used since 1986 for screening blood donations in the United States and several other countries. The advantages of p24 antigen testing are that it has high specificity and it is relatively inexpensive. Other studies have found this test to have an even higher specificity than the 99.5% we found in the present study. In studies of blood bank donors, the specificity was 99.96% or higher [11,26]. It is possible that other infections for which persons might present with suspected PHI could elevate the false-positive rate of p24 antigen testing. Although the specificity we found was slightly lower than in prior data, our results provide reassurance that p24 antigen testing has a high specificity in the setting of suspected PHI.
The shortcoming of p24 antigen testing in our study was that the sensitivity was only 79%. All of the false-negative tests occurred in persons who were reactive on an third-generation EIA test, and four of the six occurred in persons with indeterminate HIV-1 antibody tests by Western blot. This suggests that the problem with false-negative tests occurred in patients who were beginning to produce antibodies to HIV but did not have levels that achieved positive results on a standard HIV antibody testing algorithm. This indicates that there is a window period in some acutely HIV-infected patients in which p24 antigen becomes undetectable as a result of immune system activity before antibody tests are positive, similar to the window period for hepatitis B surface antibody and antigen testing in acute hepatitis B. The duration of this window period will depend on the sensitivity of the antibody test employed; using a more sensitive third-generation antibody test eliminated this window period in all our patients. Testing early in the period after onset of symptoms is also likely to eliminate problems with this window period in detection of p24 antigen. In a prior report of 20 patients tested within a week of onset of symptoms of PHI, all had detectable p24 antigen, but p24 antigen was no longer detectable within 3 weeks of onset of symptoms . The sensitivity of p24 antigen testing for persons beginning to produce antibodies could have been increased by performing acid dissociation prior to testing. However, this would increase the cost of testing and also reduces the sensitivity of p24 antigen testing to smaller amounts of virus during the initial rise in viremia.
In contrast to the p24 antigen test in which the specificity of the test remained high in the setting of suspected PHI, the specificity of the third-generation EIA test appeared to fall. In other studies, the specificity of the third-generation EIA test has been reported to be over 99%, while in our study, the specificity was only 97%. This loss in specificity may have been caused by other viral infections, which could have led to antibodies that cross-reacted on this IgM antibody-sensitive test. Other tests capable of detecting IgM antibodies have also had problems with specificity in the setting of acute infections.
HIV-1 RNA tests were highly sensitive for PHI. Individuals presenting with PHI symptoms typically had high HIV-1 RNA levels, with nearly all having > 10 000 copies/ml. These results are consistent with those of a prior study of 17 patients tested within 17 days after the onset of symptoms, in which all had HIV-1 RNA levels > 10 000 copies/ml . Participants with HIV-1 RNA levels < 10 000 copies/ml were in the resolution phase of their acute symptoms. The major shortcoming of conventional HIV-1 RNA tests was imperfect specificity. These tests have been designed to be as sensitive as possible for monitoring response to treatment for persons with known HIV infection and were not originally designed for diagnostic use for PHI. They are vulnerable to false-positive readings from low levels of contamination. Other studies have also suggested that false-positive results occur in persons who are not HIV infected [25,28]. It is important to note that all of our false-positive tests were < 2500 copies/ml HIV-1 RNA, while all the true positive tests were above this threshold. To be conservative, this suggests that HIV-1 RNA levels < 5000 copies/ml should be regarded as potential false-positive tests in the setting of suspected PHI.
HIV-1 DNA testing is an alternative to HIV-1 RNA testing for diagnosis of primary HIV infection but this was not assessed in the present study. Because HIV-1 RNA levels are higher than proviral DNA levels in early infection, DNA testing may be less sensitive. HIV-1 DNA testing uses nucleic acid amplification techniques, and it is likely to have similar problems with specificity, as minor contamination can produce false-positive results. HIV-1 RNA testing has the advantage that commercial kits are widely available in clinical settings, and the quantification of positive results provides clinically useful information.
Our results suggest that it may be useful to develop specific PHI screening test algorithms for clinical settings that would aim to reduce the risk of delivering inaccurate results to patients. One such algorithm is suggested in Fig. 3. In this algorithm, patients with suspected PHI on the basis of risk of HIV exposure and symptoms compatible with PHI are initially tested with a conventional HIV-1 antibody test and an HIV-1 RNA test. In the event that the HIV-1 antibody test is non-reactive and the HIV-1 RNA test shows detectable virus, a second test would be run to confirm the HIV-1 RNA result on a duplicate specimen. This would be particularly important if results are < 5000 copies/ml HIV-1 RNA but would provide a safeguard for tests with higher copy numbers. Results that were detectable on repeat testing but were < 5000 copies/ml might be reported as indeterminate. An antigen sandwich EIA test and/or a p24 antigen test could be used to provide further confirmation of PHI. Even with such an algorithm, clinicians need to be aware that the specificity of diagnostic testing for PHI is lower than the extremely high specificity of HIV-1 antibody tests, and that follow-up testing to confirm subsequent antibody seroconversion is desirable to provide final confirmation of the diagnosis.
The authors thank Chiron Corporation, Bager Diagnostics, Roche Molecular Systems, and Gen-Probe for providing assays used in this study.
1. Rosenberg ES, Billingsley JM, Caliendo AM. et al
. Vigorous HIV-1–specific CD4+ T cell responses associated with control of viremia. Science 1997, 278: 1447–1450.
2. Rosenberg E, Altfeld M, Poon S. et al
. Immune control of HIV-1 after early treatment of acute infection
. Nature 2000, 407: 523–526.
3. Cates W, Jr, Chesney MA, Cohen MS. Primary HIV infection – a public health opportunity. Am J Pub Health 1997, 87: 1928–1930.
4. Quinn TC, Wawer MJ, Sewankambo N. et al
. Viral load and heterosexual transmission of human immunodeficiency virus type 1. Rakai Project Study Group.
N Eng J Med 2000, 342: 921–929.
5. Royce RA, Seana A, Cates W, Jr, Cohen MS. Sexual transmission of HIV. N Eng J Med 1997, 336: 1072–1078.
6. Schacker T, Collier AC, Hughes J, Shea T, Corey L. Clinical and epidemiologic features of primary HIV infection. Ann Intern Med 1996, 125: 257–264.
7. Kahn J, Walker B. Acute human immunodeficiency virus type 1 infection. N Eng J Med 1998, 39: 33–39.
8. Kinloch-de Loes S, de Saussure P, Saurat JH, Stalder H, Hirschel B, Perrin LH. Symptomatic primary infection due to human immunodeficiency virus type 1: review of 31 cases. Clin Infect Dis 1993, 17: 59–65.
9. Perrin L, Yerly S. Clinical presentation, virological features, and treatment perspectives in primary HIV-1 infection. AIDS Clin Rev
10. Busch M, Stramer SL, Kleinman, S. Evolving applications of nucleic acid amplification assays for prevention of virus transmission by blood components and derivatives.
In Applications of Molecular biology to Blood Transfusion Medicine
. Edited by Garrarty G. Bethesda, MD: American Association of Blood Banks; 1997: 123–176.
11. Janssen RS, Satten GA, Stramer SL. et al
. New testing strategy to detect early HIV-1 infection for use in incidence estimates and for clinical and prevention purposes. JAMA 1998, 280: 42–48.
12. Fleiss JL. Statistical Methods for Rates and Proportions
, 2nd edn. New York: Wiley; 1981.
13. Hahn G, Meeker W. Statistical Intervals: A Guide for Practitioners.
New York: Wiley; 1991.
14. San Francisco Department of Public Health HIV Seroepidemiology and AIDS Surveillance Section. 1999 HIV/AIDS Epidemiology Annual Report,
Vol. 2001. San Francisco: Department of Public Health; 2000.
15. Vittinghoff E, Douglas J, Judson F, McKirnan D, MacQueen K, Buchbinder SP. Per-contact risk of human immunodeficiency virus transmission between male sexual partners. Am J Epidemiol 1999, 150: 306–311.
16. Ippolito G, Puro V, De Carli G. The risk of occupational human immunodeficiency virus infection in health care workers. Italian Multicenter Study. The Italian Study Group on Occupational Risk of HIV infection.
Arch Intern Med 1993, 153: 1451–1458.
17. Padian NS, Shiboski SC, Jewell NP. The effect of number of exposures on the risk of heterosexual HIV transmission. J Infect Dis 1990, 161: 883–887.
18. Mastro TD, de Vincenzi I. Probabilities of sexual HIV-1 transmission. AIDS 1996, 10 (Suppl A): S75–S82.
19. Tokars JI, Marcus R, Culver DH. et al
. Surveillance of HIV infection and zidovudine use among health care workers after occupational exposure to HIV-infected blood. The CDC Cooperative Needlestick Surveillance Group.
Ann Intern Med 1993, 118: 913–919.
20. Lazzarin A, Saracco A, Musicco M, Nicolosi A. Man-to-woman sexual transmission of the human immunodeficiency virus. Risk factors related to sexual behavior, man's infectiousness, and woman's susceptibility. Italian Study Group on HIV Heterosexual Transmission.
Arch Intern Med 1991, 151: 2411–2416.
21. Sackett DL. A primer on the precision and accuracy of the clinical examination. JAMA 1992, 267: 2638–2644.
22. Fox R, Odaka NJ, Brookmeyer R, Polk BF. Effect of HIV antibody disclosure on subsequent sexual activity in homosexual men. AIDS 1987, 1: 241–246.
23. Bollinger RC, Brookmeyer RS, Mehendale SM. et al
. Risk factors and clinical presentation of acute primary HIV infection in India. JAMA 1997, 278: 2085–2089.
24. Lavreys L, Thompson ML, Martin HL, Jr. et al
. Primary human immunodeficiency virus type 1 infection: clinical manifestations among women in Mombasa, Kenya. Clin Infect Dis 2000, 30: 486–490.
25. Daar E, Little S, Pitt J. et al
. Diagnosis of primary HIV-1 infection. Ann Intern Med 2001, 134: 25–29.
26. Alter HJ, Epstein JS, Swenson SG. et al
. Prevalence of human immunodeficiency virus type 1 p24 antigen in US blood donors – an assessment of the efficacy of testing in donor screening. The HIV-Antigen Study Group.
N Eng J Med 1990, 323: 1312–1317.
27. Lindback S, Karlsson AC, Mittler J. et al
. Viral dynamics in primary HIV-1 infection. Karolinska Institutet Primary HIV Infection Study Group.
AIDS 2000, 14: 2283–2291.
28. Rich JD, Merriman NA, Mylonakis E. et al
. Misdiagnosis of HIV infection by HIV-1 plasma viral load testing: a case series. Ann Intern Med 1999, 130: 37–39.